An Early Attempt at a Mechanical Perpetual Motion

Here are the only two full photos I now have of an early attempt at a mechanical perpetual motion machine. It isn't entirely naïve: the operating principle was intended to ensure:—

i) that the weights would always dwell longer on the wheel's descending side, while still exerting an average of their normal moment on that side (compared with a normally-supported weight in the same position); and

ii) that each weight would turn the wheel through a greater angle while it was falling on the descending side, compared with a smaller angle when rising.

Construction



There are two axes of rotation, separated horizontally, but kept rotating in synchronism by the gear train. The left axle carries a hub to which the bases of the springs are attached, as in this photo. The right axis of rotation is the center of the system of linkages whose purpose is to keep the "structure" of the weights as shown, i.e. always with an excess of weights on the descending side.
　

Ideally, the "structure" of the weights would be as shown above, every 1/8 of a revolution.
Although the springs were made as strong compression springs, here they are acting as
cantilever springs. Their length does not change; they only bend forward or backward with respect to the hub. During each wheel revolution, each composite-mass and its associated spring-pair make exactly one cycle of forward then backward movement.

Testing

During testing, I did achieve one aim, of spinning the wheel at a high enough rotational speed for this forward and backward movement to coincide with the natural resonant frequency of the masses and springs. As I recall, that speed was quite high. I didn't expect the resonance to be very "sharp" or of "high-Q", and it wasn't. (For one thing, gravity tries to retard the mass-spring oscillations by opposing the springs' restoring force above the horizontal centerline, and it has the opposite effect below the centerline. So the masses are always being "driven" or "driving" to some extent, even at resonance).

Although this machine didn't deliver any net energy, it generally behaved mechanically better than I had thought it might, running very smoothly at resonance. (Just as well, as a sudden stop from high speed of eight 12kg weights could have caused a fair bit of damage). Also, using a stroboscope, the structure of the weights could be seen to approximate the ideal shown in the drawing, better than the "at-rest" photos would suggest.

Several years after building this physical model, I built some silux models of devices that always had far more weight on their descending side than on their ascending side. I'll discuss two of them in future posts.